CEPS Macro Determinants of Individual Income Poverty In

8/12/2019 CEPS Macro Determinants of Individual Income Poverty In

1/34

WORKING PAPERS

Working Paper No 2010-13Avril 2010

Macro Determinantsof Individual Income Poverty

in 93 Regions of Europe

Anne REINSTADLER1

Jean-Claude RAY 2

CEPS/INSTEAD, Luxembourg1Nancy University and CNRS UMR, France 2


2/34

CEPS/INSTEAD Working Papers are intended to make research ndings available and stimulate comments and discussion.They have been approved for circulation but are to be considered preliminary. They have not been edited and have not

been subject to any peer review.

The views expressed in this paper are those of the author(s) and do not necessarily re ect views of CEPS/INSTEAD.Errors and omissions are the sole responsibility of the author(s).


3/34

1

Macro Determinantsof Individual Income Poverty in 93 regions of Europe

Anne REINSTADLER 1 CEPS/INSTEAD, Luxembourg

Jean-Claude RAY2 Nancy University and CNRS UMR 7522, France

Abstract : The analysis of the at-risk-of-poverty determinants can be improved bytaking into account factors at macro (regional) level. This hypothesis has already been madein previous research, at country-level, on cross-sectional data. We use longitudinal data inthis analysis in order to get more precise estimated parameters, and we test if the regionalunemployment rate and the regional GDP affect the individual at-risk-of-poverty status. Thecountries taken into account are those present in the Statistics on Income and LivingConditions (EU-SILC) dataset.

Key words : income poverty, EU-SILC, multilevel models, longitudinal data

JE L : I32

Avril 2010

1 Corresponding author: [email protected] [email protected]

We would like to thank David Brady, Jacques Brosius, Alessio Fusco, Tony Atkinson, Eric Marlier and PhilippeVan Kerm for useful discussions led during this project. This paper has been presented at the Net-SilcInternational Conference, 25-26 March 2010, Warsaw. Of course, remaining errors are ours.


4/34

2

1. Introduction

Tackling poverty by 2010 was one of the European objectives defined by the LisbonEuropean Council in 2000. Ten years later, 2010 is the European year for combating povertyand social exclusion. Poverty continues therefore to be at the heart of social policy in mostEuropean Member States. Ideally, social policies aimed at reducing poverty need to be basedon an in-depth understanding of the underlying processes at work. A first step towards suchan understanding consists in shedding some light on the main determinants of poverty.

Early descriptive studies have checked for relationships between poverty status anddifferent characteristics taken in turn (Bradshaw, 1999; Bradubury et al., 1999 ; UNICEF,2000 ; Mejer et al., 2000). This has given some insight on the factors involved, but thesestudies have only partially allowed to understand how these factors work. Other researches(see for example Cappellari and Jenkins, 2002; Fertig and Tamm, 2007; Brady et al., 2009)have extended this initial approach by reasoning all other things being equal, checking theeffect on poverty of factors such as educational attainment, age, employment status, familystructure all of these factors having been calculated at the individual3 level. Simultaneouslyanother stream of studies (see Moller et al., 2003; Wiepking and Maas, 2005; Brady et al.,2009; Tai and Treas, 2008) has emphasized the analysis of the role of macro characteristics ina cross-national context. These analyses have shown that the macro factors could well havean effect on the poverty probability. Indeed, the generosity of social benefits (and especiallyof family benefits) proves to have a significant negative effect on the odds of poverty (seeBrady et al. 2009, Moller et al. 2003).

For all of these three types of analyses, one major improvement has consisted in

taking into account the longitudinal feature of poverty, using panel data4

. The indicator of persistent poverty5, for example, allows to figure out whether poverty is a temporary or rathera long-term phenomenon. Furthermore, developments in the econometrics of panel data haveallowed researchers to further investigate important topics such as poverty duration orunobserved heterogeneity.

However, to our knowledge, no study has yet dealt simultaneously with all EU-countries, longitudinal data and factors at both individual and macro levels using a relevantspecification. Brady et al. (2009) study the effect of macro-determinants on the probability of being poor using a GEE model6 applied to 15 EU (plus some non-EU) countries but usingcross-sectional data. In this paper we will extend this kind of analysis to 93 EU-regions (26

countries) and, contrary to previous work, we will use a longitudinal dataset (EU-SILC 2005and 2006).

This paper has the following objective: it aims at disentangling the role of micro andmacro factors in explaining the poverty status, by using detailed information about different

3 As the same poverty status is, by the European definition, affected to all individuals belonging to the samehousehold, some authors have defined the factors exclusively at that level (see for example Andriopoulou et al.,2008).4 See Ray and Jeandidier (2003) for a comprehensive review of the French literature on this subject.5 This indicator belongs to the set of common indicators for the social protection and social inclusion process

adopted by the Social Protection Committee in 2006.6 A Generalized Estimated Equations model can be used to estimate marginal or population-averaged effectstaking into account the dependence among units nested in clusters (Rabe-Hesketh and Skrondal, 2008).


5/34

3

regions in Europe7. Indeed, we would like to test if there is a genuine effect of macro factorssuch as the unemployment rate on the poverty probability, and especially if these factors canaffect the impact of individual characteristics such as the education level on this probability.

In Section 2 we present the definition of income poverty that we will apply. In Section3 we then develop the different econometric methods available to deal with the question anddata at hand. Section 4 gives a detailed description of our dataset. The results and commentsof our own model are then presented in Section 5. Final conclusions are to be found inSection 6.

2. The definition of income poverty in Europe

In Europe, poverty is officially defined in relative terms, as the percentage ofindividuals living in a household whose equivalent income is below the poverty threshold.This threshold is defined in each country (equal to 60% of the national median equivalentincome), aiming at taking into account the national income inequalities. As a consequence,two countries with very different standards of living (and thus very different medianequivalent income and different poverty thresholds) can have the same poverty rate.

Seemingly contradictory results due to this definition do not matter as long as one isaware of the conventions they are based on, and when the at-risk-of-poverty rates areinterpreted together with the threshold values. But, in our case, the main objective is to figureout to what extent the poverty status is explained by some macro factors such as theunemployment rate. It is thus necessary, in order to allow that kind of relationship to appear,

that the poverty indicator ranks the countries as the macro variables do. As a consequence,we have chosen to keep defining poverty in a relative way (60% of a certain threshold) but tocalculate a new threshold, allowing this kind of ranking. With this objective in mind, wecalculate a unique European poverty threshold by considering all individuals to belong to asame big country, which is Europe8.

By doing this, we move away from the official EU-definition of the at-risk-of-povertystatus (which is rather a measure of intra-country inequality), and we consider Europe as anintegrated entity. In the same spirit, we exclude neither Iceland nor Norway: while not part ofthe EU-27 in 2009, they could be expected to join.

7 Indeed it is between regions, and not between individuals, that some variance of macro factors can be found.This is probably also true between countries but because our data is limited with respect to the number ofavailable countries, we prefer the analysis at the regional level.8 Using the usual at-risk-of-poverty threshold (which gives similar at-risk-of-poverty rates for countries withvery different economic situations) could lead to this kind of situation: the characteristics of the countries wouldnot be relevant to explain their own at-risk-of-poverty rate because the latter is more a measure of incomeinequalities than a measure of economic performance. In other words, macro economic factors, which can be

considered from a theoretical point of view to be associated with poverty, can not be linked with the individual poverty status such as defined when using the official definition. Note that this definition of a European at-risk-of-poverty threshold is supported by Marlier et al. (2007).


6/34

4

We then determine, for each country, the fraction of individuals9 situated below thisnew European threshold (see Appendix 1, where these figures can be compared with theofficial at-risk-of-poverty rates, based on the national thresholds).

3. Methodology

Two major approaches have been used to study the determinants of poverty. Oneconsists in explaining the transitions into and out of poverty (probability of staying poor, and probability of entering poverty). The second approach focuses on the poverty status at aspecific point in time.

The first approach takes into account the initial conditions problem10 by usinglongitudinal data. This problem refers to the fact that the poverty status during the first periodmay not be exogenous because of observed and unobserved characteristics, which wouldaffect the probability of being poor afterwards. However, papers running that kind of analysisdo not introduce macro factors in the analysis. This is either because they are interested inonly one country (see Cappellari and Jenkins, 2002, 2004; Van Kerm, 2004; Buddelmeyerand Verick, 2007; Ayllon, 200811), or because the different countries are treated separately,with as many models as there are countries (Andriopoulou et al., 2008).

On the other hand, some authors estimate the probability of being poor at a specific point in time (see Wiepking and Maas, 2005; Brady et al., 2009; Tai and Treas, 2008). All ofthese authors use cross-sectional data of 22 countries from the Luxembourg Income Study(LIS) and integrate macro variables in the analysis (such as the unemployment rate or the

welfare generosity), stressing that the welfare system could play a role in allowingindividuals to escape from poverty12.

As for our own work, it focuses on the poverty status and integrates macro factors aswell. Our method is original in two ways: first, it uses longitudinal data and second, it takesinto account the fact that some variability can be found at the regional level. In fact, tworeasons have led us to do the analysis at the regional rather than at the country-level: first because the situation the individuals face (in terms of unemployment rate for example) could be very different from one region to another, within the same country, and second becausethere are more regions than countries (93 versus 26), which is better from a statistical point ofview13.

9 These figures concern working age individuals (aged 25-55) because the behaviours and thus the factors atwork can be very different for both other groups (children and retired people).10 Some authors also control for the retention probability (see for example Cappellari, 2002; Cappellari andJenkins, 2002, 2004; Ayllon, 2008). The idea is that the probability of being observed during two consecutiveyears could depend on unobserved characteristics of the individuals that should thus be controlled for.11 Other authors have used the same kind of models, on related subjects but not poverty: Stewart and Swaffield(1999) and Cappellari (2004) on earning, Poggi (2007) on social exclusion persistence.12 Moller et al. (2003) work on these data as well, but at the macro level. Indeed, they link the national at-risk-of-poverty rate to macro variables such as GDP or the employment rate in the agricultural sector. Their sampleis quite small (61 observations, nested in 14 countries).13 While having 93 higher-level groups is technically much better than using only 26 higher-level groups,

especially as far as variances/covariances estimation is concerned, a problem remains here: regions are nestedwithin countries. This suggests to move from a three-level analysis (observations over time nested withinindividuals, themselves nested within regions) to a four-level analysis (adding the country-level at the top of the


7/34

5

In other words, we estimate a model of poverty probability, using two years ofobservations for each individual (in order to increase the estimation accuracy). Some of theseindividuals live in the same region14. From an econometric point of view, this data setupleads to a problem concerning the independence of observations: individuals being observedover two years and/or living in the same region share their own time-invariant characteristicsand/or the characteristics of the area and can therefore no longer be considered to beindependent. As a consequence, using traditional techniques would give consistent estimates but heavily15 under-estimated standard errors.

In order to cope with that statistical problem, we have chosen one of the manyavailable techniques: we run a multilevel model, which treats the upper levels (the individualsand the regions here) not as unique entities but as units primarily characterised by factorscalculated at their level (e.g. characteristics of the individual, or the unemployment rate of theregion). These models explicitly take into account the hierarchical structure of the data,thereby allowing us to analyse first to measure, then to explain the fraction of thevariability of the poverty rate which is attached to each (nested) level. Contrary to the fixedeffects models16, multilevel models make use of the between variance, and are thereforeespecially useful when this variance is quite high. Some authors have already stressed that theuse of this kind of models would be relevant in this framework (Cappellari and Jenkins 2004;Brady et al. 2009) but they have underlined the complexity of these models, whoseconvergence status is often out of reach.

The model we estimate is a binary logistic regression, where the probability of beingat-risk-of-poverty is explained. This multilevel model takes into account three levels: time(measured in years), individuals and regions. It can be written as follows in its structuralform17:

= 0 + 1 1 + 2 2 0 = 0 + 1 0 + 0 1 = 0 + 1 1 + 1

hierarchy). But models with four levels and so many observations prove to have convergence problems and canthus not be estimated. We have therefore chosen to restrict our analysis to three levels, and to consider the thirdas being either the regions (in which case the country variable is introduced in the model as an explanatoryvariable) or the countries (the regional level being then put aside see Appendix 2).14 Some of them also live in the same household. Ideally, this fact should be taken into account as well, butagain it would increase the complexity of the analysis: we would have to deal with 3 sources of non

independence of the observations (individuals observed several years, living in the same household, and in thesame region). In this paper, we cope with two of them.15 According to what Angrist and Pischke(2009) name the M oulton factor (p. 310), we calculate to whatextent the macro factors regression coefficients would be overestimated by ignoring intraregional correlation.We use the general formula 8.2.5 (ibidem p. 311) which allows for various cluster sizes. With an intra-classcorrelation coefficient (given by the empty model see Appendix 6) of 0.551, a (non weighted) averageregional sample size of 3206.20 individuals, a (non weighted) variance of regional sample size amounting to15,550,017.33 and an intra-class correlation of macro factors equal to 1 by definition, we get 66.63. Note thatthis very large impact of clustering on standard errors is here due to the conjunction of big discrepancies between the size of regions (from 19 to 19941) and a high intra-class correlation coefficient (0.551). Even if theMoulton formula should be considered here only as a very rough approximation (because this formula ignoresweighting and repeated observations), it clearly suggests that clustering effects must be taken into account.16 Yet these models have a strong advantage: they control for group-invariant factors, measured or unmeasured.

But this advantage has a price: the inability to estimate regression coefficients for these group-invariant factors,and thus to allow the analyst to conclude in terms of the effect of these factors.17 The notation we use here is the same as Snijders and Bosker's (2004).


8/34

6

where:i indexes time j indexes the individualsk indexes the regions

is the probability of being at-risk-of-poverty1 is a vector of independent factors defined at the individual/household level whose

effects are assumed to be random2 is a vector of independent factors defined at the individual/household level whose

effects are assumed to be fixed0 is a vector of independent factors defined at the regional level, which are supposed to

have an impact on the average in region k1 is a vector of independent factors defined at the regional level, which are supposed to

moderate the effect of the1 on 0 is a random intercept1 is a vector of random slopes2 is a vector of fixed slopes0 measures the average value of across regions, when each independent variable is 01 measures the impact of0 on 0 measures the average impact, across regions, of1 on , when each 1 is 01 measures the impact of1 on the effect of the 1 on 0 and 1 are error terms assumed to follow a multinormal distribution(0,0; ) , being

the variance-covariance matrix18.

The reduced form is thus:

= 0 + 1 0 + 0 1 + 1 1 1 + 2 2 + 0 + 1 1 This formula refers to a random slope model, meaning that the intercept and at least

one of the explanatory variables have a random coefficient.

4. Data

The EU-SILC longitudinal dataset provides information at both individual andhousehold levels and covers at most 5 years (from 2003 to 2007) depending on the country:data are not available for some countries in 2003, 2004 and 2007. Had we used all five wavesto calculate the poverty threshold as we define it (i.e. at the European level), it would haveincreased or decreased over time just because some countries (e.g. Germany) are absent forsome years it means, without any link with the economic situation. As a consequence, wework with data from two waves only, 2005 and 2006, where all 26 countries are present.

The unit of analysis is the individual: as stated by an OECD report (2001), this is theusual choice for poverty analysis with longitudinal data because individuals can be followed

18

In our main model, we specify an unstructured form of the variance-covariance matrix (allowing covariances between random effects to be non zero) because the covariance between the error terms of the intercept and thenumber of employed people in the household appears to be highly significant.


9/34

7

over time whereas households cannot19. The sample contains 131 891 working age adults(25-55) for the first wave, and 166 379 for the second20, split between 26 countries (seeAppendix 3). These countries21 are in turn divided into 93 regions (see Appendix 4).

The explanatory variables have been chosen in order to control for differentdeterminants of the poverty status. Some are related to the demographic characteristics of thehousehold22 (number of children and number of adults), others to the labour market(presence of at least one adult with an upper level of education, number of employed people),others still to the health status (presence23 of at least one adult with chronic disease, orhampered by illness in his/her daily activities). Two additional variables are measured at theregional level: the GDP and the unemployment rate24.

Some descriptive statistics, for both the whole sample and each country, can be foundin Appendix 5 (but not for each region: as there are 93 regions25, it proves not to be sensibleto show the descriptive statistics for each of them).

Beyond the usual hypotheses concerning all the control variables26, we make twofurther ones concerning our variables of interest: first we assume that the negative effect ofthe level of education on the probability of being at-risk-of-poverty could be weaker in richerareas (where the probability of being poor is quite low, whatever the level of education).Second we assume that the negative effect of the number of employed people in the

19 As usual, the poverty threshold is calculated at the individual level but the poverty status is defined at thehousehold level.20 Note that the sample is not balanced: 41 % of the individuals have only one observation (only 5% inDenmark, up to 59% in Czech Republic).21

The latest release of SILC longitudinal data (August 2009) contains only 22 countries, Germany, Ireland,Greece and Denmark being absent. But we wanted both to work on this release, the data of which have beencleared of previous problems, and to keep these four countries in the analysis. We have therefore added to these22 countries the other 4 from the previous release (March 2009).22 All of the explanatory variables, apart from gender and age, are calculated at the household level, because the poverty status is defined at that level.23 Several variables have originally been defined at the individual level (e.g. having a chronic disease). In orderto have all variables defined at the same (household) level, we have tried to build aggregated variables such asthe number of adults in the household suffering from a chronic disease. Unfortunately, this information wasavailable only for one individual per household in the register countries (which use information fromadministrative datasets when available and interview only one individual per household for the remainingquestions to be asked). Keeping this kind of definition would thus have led to a big loss of information. As aconsequence, we have defined a much less precise indicator, such as "presence in the household of at least one

adult suffering from a chronic disease". Note that even this imprecise indicator could be ill measured in thoseregister countries, as only one household member is interviewed (and the construction of the variable wouldthen only rest on that member). The level of education is defined for all household members aged 16 or more, but the variable contains a lot of missing values in some countries (13% in Portugal, 14% in Spain and up to16% in the UK in 2005). We have thus adopted the same definition in order to construct a variable at householdlevel.24 Unfortunately, we were not able to integrate in our main model potential important macro factors such as thesocial expenditures (expressed as percentages of the GDP): unemployment compensation, public healthexpenditures, expenditures with respect to inclusion. Indeed, they are not yet known at regional level.25 In fact, 16 countries out of the 26 countries at hand only have one region, either because they are quite small(12 of them), or because the region code is not available in the dataset (4 of them).26 Note that income (and thus the poverty status) and individual/household demographic characteristics have not been measured at the same time: income refers to the year prior to the survey, whereas demographic

characteristics to the time of the survey. We could have dealt with this by lagging all non-income variables (e.g.demographic characteristics given for year 2006 should be linked to the income declared in 2007), yet at the price of losing some countries since not all of them have available data for 2007.


10/34

8

household on the probability of being at risk of poverty could be attenuated when theunemployment rate is high due to the downward pressure on wages.

5. Results and Comments

The results of the model are shown in Table 1. All our analyses (descriptive andeconometric) use weighted data27.

27 Moon and Stotsky (1993) state that, if the data come from a stratified and clustered random sampling, it isreasonable to treat the sample as a simple random sample, thus ignoring weights. And Poggi (2007) states that itis more efficient, from an econometrical point of view, not to weigh the data. That way of doing has been

adopted by studies on panel data (as stated by Ayllon, 2008, or Andriopoulou et al., 2008). But the question ofweighting the data is still open). We have chosen to weigh them, as the SILC dataset is known not to berepresentative of the population.


11/34

9

Table 1. Probability of being at-risk-of-poverty in 93 European regions. Estimation with a multilevel modelParameter estimate Standard error Odds Ratios

Variables 29 interceptwomanage centered (around the average : 41.27)age centered squaredchronic disease in the householdactivity hampered by disease in the householdupper education level in the householdnumber of children in the householdnumber of children in the household squarednumber of adults in the householdnumber of adults in the household squarednumber of employed people in the householdregional annual GDP per capita (in 10 Euros)regional unemployment rate (expressed in %)upper education level * regional GDP per capitanumber of employed people * regional unemployment rate

0,26250,015780,002428 *0,000453 ***

-0,1214 ***0,1612 ***

-1,5241 ***0,3555 ***

-0,0127 *-0,2607 ***0,08983 ***

-1,2964 ***-0,04818 ***0,021030,02427 **0,01392 *

0,38060,01390,0008480,0001050,016920,017540,15680,015320,0042980,027860,0046590,073990,0073530,011220,0066890,005996

1,0161,0021,0000,8861,1750,2181,4270,9870,7701,0940,2740,9531,021

wave 2005wave 2006

-0,2055 ***ref.

0,01488ref.

0,814ref.

country BEcountry CZcountry DKcountry DEcountry EEcountry IEcountry ELcountry EScountry FRcountry ITcountry CYcountry LVcountry LTcountry LU

country HUcountry NLcountry ATcountry PLcountry PTcountry SIcountry SKcountry FIcountry SEcountry UKcountry IScountry NO

-0,32081,9401 ***-0,6816-0,27403,1916 ***

-0,24530,9511 *0,9605 *

-0,28640,2918

-0,22653,5168 ***3,7270 ***

-0,3031

3,0947 ***-0,2287-0,32972,5769 ***1,4376 **0,17963,2642 ***

-0,5984-0,3938

ref.-0,4673-0,4939

0,36160,32780,44190,42520,44600,44680,34310,31150,31070,33110,48890,44760,44450,7205

0,35810,42670,35660,34750,43210,45160,44150,35360,4318

ref.0,66920,4646

0,7266,9600,5060,760

24,3280,7822,5892,6130,7511,3390,797

33,67841,5560,739

22,0810,7960,719

13,1574,2101,197

26,1600,5500,675

ref.0,6270,610

Regional-level error terms variancesinterceptupper education level in the householdnumber of employed people in the household

Regional-level error terms covariancesCOV (intercept, upper education level in the household)COV (intercept, number of employed people in the HH)COV (upper education level, nb. of employed people in the HH)Other parametersRho coefficient of AR(1)ResidualFit measure: -2 Log Pseudo Likelihood

0,38180,20770,2198

-0,04609-0,2342-0,0437

0,31380,9936

1832580

0,082750,044860,03756

0,042810,050930,03109

0,0028170,002727

Source: EU-SILC data, longitudinal file30, 1.08.2009 UDB release, authors' computations.Level of significance for independent variable coefficients: *: p-value < 0.05; **: p-value < 0.01; ***: p-value < 0.001

28 We have used the SAS GLIMMIX command. Useful SAS code examples can be found in Allison (2008) and

were kindly given by David Brady.29 See appendix 5 for a description of the variables.30 22 countries from this release plus 4 countries from the March release (see above).


12/34

10

Concerning the error terms, SAS/PROC GLIMMIX does not offer a statistical testindicating the level of significance of the variances and covariances of the error terms. Butcompared to their standard errors, the estimated variances are quite high, which suggests theirhigh level of significance. This in turn justifies on the one hand the choice of the multilevelmodel, and on the other hand our choice of allowing these variables to have random ratherthan fixed coefficients. Looking at the empty model (see Appendix 6), we can see that theintra-class correlation (calculated according to the second formula given by Snijders andBosker, 1999, page 224) is equal to 0.55, meaning that the between-variance is substantial.

Let us examine now the effects of our variables of interest. Recall that our objective isto measure the specific effect of the regional GDP per inhabitant and the regionalunemployment rate on the probability of being at-risk-of-poverty. This effect could be eitherdirect or indirect since these macro factors can act through other individual variables on the probability of being poor (such as the education level and the number of employed people inthe household).

As expected, the regional GDP per capita has a strong (and highly significant) directnegative effect on the risk of poverty: for individuals living in households where nobody hasan upper education level, the odds of being poor (probability of being poor divided by probability of not being poor) decreases by 4.7% (1-0.953=0.047) for an increase of annualGDP per capita by a 1000 Euros. This direct effect is supplemented by an indirect effect: theregional GDP per capita moderates the negative impact of upper education on the povertyrisk. In fact, in the average region in terms of GDP per capita (about 24260 Euros/year), the presence of an adult with upper education level decreases the poverty odds by 61% (oddsratio = 0.3931). In a rich region such as Luxembourg (GDP per capita = 60150 Euros/year), itdecreases the odds by only 6.2%; in a quite disadvantaged region like Estonia (GDP percapita = 14547 Euros/year) it decreases the odds by 69 %.

For individuals living in households without any employed people, an additional percentage point of the regional unemployment rate increases by 2.1% the poverty odds, butthis effect is only slightly statistically significant (p-value=6%) ceteris paribus (especiallywhen GDP per capita is controlled for). But there is an indirect effect of the regionalunemployment rate on the poverty risk, even if rather small: in the average region in terms ofthe unemployment rate (unemployment rate = 8.52 %), the presence of an additionalemployed individual decreases the poverty odds by 69%, when controlling for the number ofadults in the household (odds ratio = 0.3132). When the unemployment rate is much lower,

such as in Ireland (unemployment rate 4.35%), it decreases these odds by 71%, and by 66%when the unemployment rate is quite high (for example in Slovenia - unemployment rate14.85%).

To summarize, the moderating effect of the regional unemployment rate on the impacton poverty risk of the number of employed people does exist but it is marginal. By contrast,the moderating effect of the regional GDP per capita on the impact on poverty risk of the presence of highly educated people is quite large.

31 Odds ratio = 0.39 = exp(-1.5241 + 0.02427*24.260)32 Odds ratio = 0.31 = exp(-1.2964 + 0.01392*8.52)


13/34

11

Besides and not surprisingly (given the sample size), almost all control variables havean effect on the poverty probability, except for gender 33: women do not have a higher risk of being poor than men. This can easily be explained by the fact that the poverty status is ahousehold characteristic, which can hardly be influenced by a strictly individual feature34.One interesting question is the extent to which these control variables have an impact on the poverty risk (even if our study focuses on the possible impact of macro determinants on theeffect some factors of interest can have on the poverty risk):

- if the activity of at least one household member is hampered by disease, the odds of being poor increase by 18%

- controlling for especially the fact that the activity of at least one household memberis hampered (or not) by disease, the chronic character of this disease decreases by 11% theodds of being poor (this could be a consequence of the social benefits people with chronicdiseases are entitled to)

- the number of children in the household has a slightly concave effect on the odds of being poor: as the number of children increases, the effect of an additional child decreases progressively up to a value (14 children) lying beyond the observed maximum in the sample(12 children). But the effect of each additional child remains substantial. As an example,while a first child increases by 41% the odds of being poor, a fourth child still increases them by 31%

- the number of adults in the household also proves to have a clear non linear effect onthe poverty risk: ceteris paribus (and especially when controlling for the number of employed people), the odds of being poor are virtually the same if there are one or two adults in thehousehold, but they increase by 21% with the third adult and by 45% with the fourth one.

6. Conclusion

Analysing the determinants of the monetary poverty probability has already beenattempted by many studies. But few of them have simultaneously used panel data, consideredfactors at the macro level, and used the right techniques to deal with all these elements.

As for our results, they show that both the regional GDP per capita and the regional

unemployment rate do have an effect on poverty risk.

33 Since the poverty status is defined at the household level, it could be considered as irrelevant to introducevariables measured at individual level such as age and gender. We have added these two variables to our modelfor comparison's sake, as almost all studies do this as well. As far as age is concerned, it could be argued that,even though age is, like gender, a factor measured at the individual level, it might have some signification as anhousehold characteristic: due to frequent endogamy, the age of one adult belonging to the household offerssome clue about the age of other adults in the household, at least for non single adult households. With theselimitations in mind, we can interpret the (highly significant) quadratic effect of age: ceteris paribus, the povertyrisk is first decreasing more than linearly with age, reaching a minimum at age 44, and then increasing morethan linearly. For example, at age 25, an additional year results in a decrease of the poverty risk by 1.2%; and at

age 50, an additional year of age results in an increase of 1.1%.34 This is the very reason why the different common indicators agreed upon in the context of the OMC on SocialProtection and Social Inclusion which focus on gender differences are calculated on single person households.


14/34

12

In terms of economic and social policy implications, it means that:

- policies oriented towards higher economic growth rates in disadvantaged Europeanregions are able to alleviate the risk of poverty even if poverty is defined in relative terms

- this kind of economic policies, if successful in its effort to sustain the economicwell-being of families in poor regions, will, as a side-effect, diminish the anti-poverty effectof the presence in the household of higher educated people. We suspect that this indirecteffect is associated to the choice of defining poverty as a relative concept a European view,which countries like the US do not share

- as for the regional unemployment rate, its direct positive impact on the poverty riskis essentially a confirmation of what could be expected and of what is already known, even ifthe weakness of this effect is quite surprising. Still more surprising is the fact that theregional unemployment rate does not moderate to a large extent the impact on the povertyrisk of the number of employed people in the household. This implies that policies aimed atcombating high regional unemployment rates will, as such, unfortunately not lower to a largeextent the regional poverty rates.

However, our analysis faces two types of limitations. The first one results frommethodological choices we have made, the second is due to the data.

First, because we needed an indicator differentiating and thus ranking the 93 regionsin terms of poverty rates, we have made use of a European poverty threshold, which has proven quite relevant in terms of ability to estimate the econometric model. However, precisely because the European regions are quite dispersed around the average at-risk-of- poverty rate, we were not able to check the consistency of the results in using alternativemeasures of the European threshold35 (such as 50% or 70% of the European medianequivalent income).

Concerning then the data, and with the results at country-level in mind, we wouldhave liked to test the effect of other macro characteristics at the regional level, such as theexpenses in unemployment or social benefits (expressed in percentage of the GDP). But thesewere not available at the regional level. In future analyses, we would thus be interested inadding some variables of that kind, once they are available. And, as for the regions, thevariable defined in the EU-SILC dataset is missing in some countries (even in quite bigcountries such as Germany and the United Kingdom). In order to keep these countries in the

analysis, we have defined each of them as a unique and quite large region.

35 In fact, with a poverty threshold equal to 60% of the European median equivalent income, the at-risk-of-

povery rates of the different countries range between 1% and 82% in 2005, and between 2% and 75% in 2006.Changing this threshold for a lower (higher) one would lead to even lower (higher) rates in the richest (poorest)countries, making the analysis impossible to run.


15/34

13

References

ALLISON Paul D. (2008), Longitudinal Data Analysis Using SAS, October, 100 pages.

ANDRIOPOULOU Eirini, TSAKLOGLOU Panos (2008), "Once poor, always poor? Do initialconditions matter? Evidence from the ECHP", IRISS Conference, Oct., 33 pages.

ANGRIST Joshua D., PISCHKE Jrn-Steffen (2009), Mostly Harmless Econometrics. AnEmpiricists Companion , Princeton University Press, 373 p.

AYLLON Sara (2008), "Modelling poverty transitions in Spain: Do attrition and initial conditionsreally matter?", IRISS Working Paper Series, 2008-08, Oct., 36 pages.

BRADBURY Bruce, JNTTI Markus (1999) Child Poverty across Industrialised Nations,Innocenti Occasional Papers, Economic and Social Policy Series, UNICEF, n71.

BRADSHAW Jonathan (1999), "Child Poverty in Comparative Perspective", European Journal ofSocial Security , Vol.1, pp. 383-406.

BRADY David, FULLERTON Andrew S., MOREN-CROSS Jennifer (2009), "Putting Poverty inPolitical Context: A Multi-Level Analysis of Adult Poverty Across 18 Affluent Countries",Social

Forces , Vol. 88(1), September, pp. 271-300.

BUDDELMEYER Hielke, VERICK Sher (2007), "Understanding the Drivers of Poverty Dynamicsin Australian Households", IZA DP No. 2827, 29 pages.

CAPPELLARI Lorenzo (2004), "Earnings Mobility Among Italian Low Paid Workers", Journal of Population Economics , Vol. 20(3), pp. 465-482.

CAPPELLARI Lorenzo, JENKINS Stephen P. (2002a), "Who Stays Poor? Who Becomes Poor?Evidence from the British Household Panel Survey",The Economic Journal , Vol. 112, No. 478,Conference Papers (Mar., 2002), pp. C60-C67.

CAPPELLARI Lorenzo, JENKINS Stephen P. (2002b), "Modelling Low Income Transitions", ISER Working Papers Number 2002-8 , 42 pages.

CAPPELLARI Lorenzo, JENKINS Stephen P. (2004), "Modelling Low Income Transitions", Journal of Applied Econometrics, Vol. 19, pp. 593 610.

FERTIG Michael, TAMM Marcus (2007), "Always Poor or Never Poor and Nothing in Between?

Duration of Child Poverty in Germany", IZA Discussion Paper N 2645, February, 29 pages.MARLIER Eric, ATKINSON A.B., CANTILLON Bea, NOLAN Brian (2007), The EU andSocial Inclusion Facing the challenges, The Policy Press, The University of Bristol, 303 pages.

MEJER Lene, SIERMANN Clemens (2000), "La pauvret montaire dans l'Union europenne : lasituation des enfants, les diffrences entre les sexes et l'cart de pauvret", Eurostat,Statistiques enbref , Population et Conditions sociales, 12/2000, 7 pages.

MOLLER Stephanie, HUBER Evelyne, STEPHENS John D., BRADLEY David, NIELSENFranois (2003), "Determinants of Relative Poverty in Advanced Capitalist Democracies", AmericanSociological Review , Vol. 68, No. 1, February, pp. 22-51.


16/34

14

MOON Choon-Geol, STOTSKY Janet G. (1993), "The Effect of Rent Control on Housing QualityChange: A Longitudinal Analysis",The Journal of Political Economy , Vol. 101, No. 6, December, pp.1114-1148.

OECD (2001), "When Money is Tight: Poverty Dynamics in OECD Countries", in EmploymentOutlook , Chapter 2, pp. 37-87.

POGGI Ambra (2007), "Does persistence of social exclusion exist in Spain? ", Journal of Economic Inequality , Vol. 5, pp. 53-72.

RAY Jean-Claude , JEANDIDIER Bruno (2003), "Depuis une dcennie, que nous ont appris lesdonnes longitudinales propos de la pauvret en France ? Une premire synthse", Communicationaux Xes Journes du Longitudinal, Caen, mai, 61 pages.

SNIJDERS Tom A.B., BOSKER Roel J. (2004), Multilevel Analysis. An introduction to basic andadvanced multilevel modeling , Sage Publications, 266 pages.

STEWART Mark B. , SWAFFIELD Joanna K. (1999), "Low Pay Dynamics and TransitionProbabilities", Economica , Vol. 66, pp. 23-42.

TAI Tsui-o, TREAS Judith (2008), " Poverty, Household Composition, and Welfare States: AMulti-level Analysis of 22 Countries", Luxembourg Income Study Working Paper Series, WorkingPaper No. 492, September, 33 pages.

UNICEF (2000), "A League Table of Child Poverty in Rich Nations", Innocenti Report Card, Issuen1, June, 28 pages.

VAN KERM Philippe (2004), "Une valuation conomtrique des flux vers et hors de la pauvret enBelgique", IRISS Working Paper Series, 2004-04, 17 pages.

WIEPKING Pamala , MAAS Ineke (2005), "Gender Differences in Poverty: A Cross-NationalStudy", European Sociological Review , Vol. 21(3), July, pp. 187-200.


17/34

15

Appendix 1. At-risk-of-poverty rates of working aged adults (25-55) in Europe

At-risk-of-poverty rates of working age adults (25-55) in Europe(European threshold = 60% of the European median equivalentincome)

Country 2005 2006ATBECYCZDEDKEEESFIFRELHU

IEISITLTLULV NL NOPLPTSESISK

UK

567

4273

702058

2473

8314791

8163

74425

1377

7

686

4163

641969

2669

8317752

7433

73405

1372

8Source: EU-SILC data, longitudinal file, 1.08.2009 UDB release,authors' computations.Reading note: with the European poverty threshold calculated for thewhole population36, 5% of individuals aged 25-55 in Austria were at-risk-of-poverty in 2005.

We can notice that the countries face very different situations in terms of the at-risk-of-poverty rate, a conclusion that cannot be drawn from the figures shown in the table below(based on national poverty thresholds).

36

The European poverty threshold is calculated by taking into account all individuals living in the 26 countriesunder study. In other words, children and elderly people are not excluded from this calculation, even if they arenot kept in the analyses afterwards.


18/34

16

At-risk-of-poverty rates of working aged adults (25-55) in Europe(national thresholds)

Country 2005 2006ATBECY

CZDEDKEEESFIFRELHUIEISITLTLULV NL NOPLPTSESISKUK

111110

10119

16168

101614149

16191318108

21157

101314

111110

9129

15159

111715148

1718141988

1915109

1114

Source: EU-SILC data, cross-sectional file, 1.08.2009 UDB release.Reading note: with the poverty threshold calculated at national level,11% of individuals aged 25-55 in Austria were at-risk-of-poverty in2005.

These official figures are based on the cross-sectional file, which means that allindividuals are taken into account (whereas the longitudinal file concerns only those presentat least two years).


19/34

17

Appendix 2. Poverty probability determinants at country-levelResponse Profile

Ordered TotalValue pov_indicator Frequency

1 1 722182 0 211983

The GLIMMIX procedure is modeling the probability that pov_indicator='1'.Dimensions

G-side Cov. Parameters 3R-side Cov. Parameters 2Columns in X 21Columns in Z per Subject 3Subjects (Blocks in V) 26Max Obs per Subject 30703

Optimization InformationOptimization Technique Newton-RaphsonParameters in Optimization 4Lower Boundaries 4Upper Boundaries 1Fixed Effects ProfiledResidual Variance ProfiledStarting From GLM estimates

/ 32 iterations

Convergence criterion (PCONV=1.11022E-8) satisfied.Fit Statistics-2 Res Log Pseudo-Likelihood 1834383Generalized Chi-Square 291051.3Gener. Chi-Square / DF 1.02

Covariance Parameter EstimatesStandard

Cov Parm Subject Estimate ErrorIntercept country 0.3242 0.1091uppereducHH country 0.2754 0.08766nbemployedHH country 0.1406 0.04498AR(1) ID_unique_UE27(country) 0.3201 0.002776Residual 1.0242 0.002811

Asymptotic Covariance Matrix of Covariance Parameter EstimatesCov Parm Subject CovP1 CovP2 CovP3 CovP4 CovP5Intercept country 0.01190 0.000089 -0.00019 -6.96E-7 -5.7E-7uppereducHH country 0.000089 0.007684 0.000032 -6.59E-7 -3.03E-7nbemployedHH country -0.00019 0.000032 0.002024 3.937E-7 2.309E-8AR(1) ID_unique_UE27(country) -6.96E-7 -6.59E-7 3.937E-7 7.707E-6 2.001E-6Residual -5.7E-7 -3.03E-7 2.309E-8 2.001E-6 7.903E-6

Asymptotic Correlation Matrix of Covariance Parameter EstimatesCov Parm Subject CovP1 CovP2 CovP3 CovP4 CovP5Intercept country 1.0000 0.009333 -0.03776 -0.00230 -0.00186uppereducHH country 0.009333 1.0000 0.008229 -0.00271 -0.00123nbemployedHH country -0.03776 0.008229 1.0000 0.003152 0.000183AR(1) ID_unique_UE27(country) -0.00230 -0.00271 0.003152 1.0000 0.2564Residual -0.00186 -0.00123 0.000183 0.2564 1.0000

Solutions for Fixed EffectsStandard

Effect wave Estimate Error DF t Value Pr > |t|Intercept 1.4742 0.4733 25 3.11 0.0046wave 2005 -0.1846 0.02712 284E3 -6.81


20/34

18

Appendix 3. Sample size, by country and year

Number of individuals aged 25-55, in each country, for each year (sample size,

unweighted cases)Country 2005 2006ATBECYCZDEDKEEESFIFRELHUIEISITLTLULV NL NOPLPTSESISKUK

3882240823574391

10529354926658472402158764642435328721584

1295525334322270485744088

1062726713541867134476224

5469400233737367928534514269

11501552873244211660919922243

1804837134601343892873926

1464436624751

1127148987594

Source: EU-SILC data, longitudinal file, 1.08.2009 UDB release, authors' computations.


21/34

19

Appendix 4. List of available regions37 in the dataset

AustriaAT1 Oststerreich

AT2 SdsterreichAT3 Weststerreich

BelgiumBE1 Rgion de Bruxelles-Capitale /

Brussels Hoofdstedelijk GewestBE2 Vlaams GewestBE3 Rgion Wallonne

Republic of CyprusCY0 Kypros / Kibris

Czech Republic

CZ01 PrahaCZ02 Stredni CechyCZ03 JihozapadCZ04 SeverozapadCZ05 SeverovychodCZ06 JihovychodCZ07 Stredni MoravaCZ08 Moravskoslezsko

GermanyThe region variable is not available in thelongitudinal dataset for Germany.DE Germany

DenmarkDK0 Denmark

EstoniaEE0 Estonia

SpainES11 GaliciaES12 Principado de AsturiasES13 CantabriaES21 Pas VascoES22 Comunidad Foral de NavarraES23 La RiojaES24 AragnES30 Comunidad de MadridES41 Castilla y LenES42 Castilla-La ManchaES43 ExtremaduraES51 CataluaES52 Comunidad ValencianaES53 Illes BalearsES61 Andaluca

37

Details from EU-SILC documentation andhttp://ec.europa.eu/eurostat/ramon/nuts/codelist_en.cf m?list=nuts

ES62 Regin de MurciaES63 Ciudad Autnoma de Ceuta

ES64 Ciudad Autnoma de MelillaES70 Canarias

FinlandFI13 It-SuomiFI18 Etel-SuomiFI19 Lnsi-SuomiFI1A Pohjois-Suomi

FranceFR10 le de FranceFR21 Champagne-ArdenneFR22 Picardie

FR23 Haute-NormandieFR24 CentreFR25 Basse-NormandieFR26 BourgogneFR30 Nord - Pas-de-CalaisFR41 LorraineFR42 AlsaceFR43 Franche-ComtFR51 Pays de la LoireFR52 BretagneFR53 Poitou-CharentesFR61 AquitaineFR62 Midi-PyrnesFR63 LimousinFR71 Rhne-AlpesFR72 AuvergneFR81 Languedoc-RoussillonFR82 Provence-Alpes-Cte d'AzurFR83 Corse

GreeceGR1 Voreia ElladaGR2 Kentriki ElladaGR3 AttikiGR4 Nisia Aigaiou, Kriti

HungaryHU1 Kozep-MagyarorszagHU2 DunantulHU3 Alfold Es Eszak

IrelandThe region variable is not available in thelongitudinal dataset for Ireland.IE0 Ireland

IcelandIS Iceland


22/34

20

ItalyITC Nord-OvestITD Nord-EstITE Centro (I)ITF SudITG Isole

LithuaniaLT0 Lietuva

LuxembourgLU0 Luxembourg (Grand-Duch)

LatviaLV0 Latvija

The NetherlandThe region variable is not available in thelongitudinal dataset for the Netherlands.

NL The NetherlandsNorway NO0 Norway

PolandPL1 Region CentralnyPL2 Region PoludniowyPL3 Region WschodniPL4 Region Polnocno-ZachodniPL5 Region Poludniowo-Zachodni

PL6 Region PolnocnyPortugalPT Portugal

SwedenSE Sweden

SloveniaSI Slovenia

SlovakiaSK0 Slovenska

The United KingdomThe region variable is not available in thelongitudinal dataset for UKUK United-Kingdom


23/34

21

Appendix 5. Description of the explanatory variables and descriptive statistics

Descriptive statistics for the whole sample

Wave N Obs Variable N NMiss Mean Std Dev Minimum Maximum---------------------------------------------------------------------------------------------

2005 131891 pov_indicator 131891 0 0.1965 0.4259 0 1.0000woman 131887 4 0.5130 0.5357 0 1.0000age_centered 131891 0 -0.6425 9.1801 -16.2665 14.7335age_centered2 131891 0 73.7935 76.2358 0.0710 264.6chronicdiseaseHH 131891 0 0.3885 0.5223 0 1.0000activityhamperedHH 131891 0 0.2931 0.4878 0 1.0000uppereducHH 126795 5096 0.3962 0.5182 0 1.0000nbchildren 131891 0 0.9077 1.1426 0 11.0000nbadultsHH 131891 0 2.3903 1.0792 1.0000 10.0000nbemployedHH 129384 2507 1.5131 0.8971 0 8.0000GDPhabnuts 131891 0 23.7538 7.1961 8.2000 57.1000unempratenuts 131891 0 8.8728 4.4381 2.5000 21.4000

2006 166379 pov_indicator 166379 0 0.2000 0.3757 0 1.0000woman 166371 8 0.5128 0.4695 0 1.0000age_centered 166379 0 -0.2041 7.9962 -16.2665 14.7335age_centered2 166379 0 72.5178 66.5149 0.0710 264.6chronicdiseaseHH 166379 0 0.3911 0.4584 0 1.0000activityhamperedHH 166379 0 0.2926 0.4273 0 1.0000uppereducHH 158892 7487 0.4059 0.4560 0 1.0000nbchildren 166379 0 0.9288 0.9917 0 12.0000nbadultsHH 166379 0 2.4239 0.9506 1.0000 11.0000nbemployedHH 162220 4159 1.6016 0.7713 0 8.0000GDPhabnuts 166379 0 24.7924 6.6164 8.7000 63.1000unempratenuts 166379 0 8.1568 3.0558 2.8000 21.0000

Description of the explanatory variables Name of the variable Label of the variable Description of the variablewoman woman EU-SILC variable RB090; woman=1 if RB090=2age_centered age centered (around the

average: 41.27)EU-SILC variable RX020, centered (age in the year prior to the survey)

chronicdiseaseHH chronic disease in thehousehold

Authors' calculations using the EU-SILC variablePH020: is there at least one household member whosuffers from a chronic disease?

activityhamperedHH activity hampered by disease inthe household

Authors' calculations using the EU-SILC variablePH030: is there at least one household memberwhose activities are hampered because of health problems?

uppereducHH upper education level in thehousehold

Authors' calculations using the EU-SILC variablePE040: is there at least one household memberwhose upper level of education is tertiary education

(PE040=5)?nbchildren number of children in thehousehold

number of children (age 0-14) in the year prior to thesurvey

nbadultsHH number of adults in thehousehold

number of adults (age 18 or more) in the year prior tothe survey

nbemployedHH number of employed people inthe household

number of employed household members (authors'calculations using the EU-SILC variable PL030 codes 1 or 2)

country country EU-SILC variable RB020wave wave EU-SILC variable RB010GDPhabnuts regional annual GDP per capita

(in 10 Euros)Information from Eurostat

unempratenuts regional unemployment rate

(expressed in %)

Information from Eurostat


24/34


25/34

23

activityhamperedHH 10529 0 0.4231 0.7786 0 1.0000uppereducHH 10364 165 0.5070 0.7878 0 1.0000nbchildren 10529 0 0.7408 1.5855 0 6.0000nbadultsHH 10529 0 2.1375 1.3513 1.0000 7.0000nbemployedHH 10529 0 1.3698 1.2661 0 5.0000GDPhabnuts 10529 0 26.3000 0 26.3000 26.3000unempratenuts 10529 0 11.1000 0 11.1000 11.1000

2006 DE 9285 pov_indicator 9285 0 0.0624 0.3895 0 1.0000

woman 9285 0 0.5655 0.7983 0 1.0000age_centered 9285 0 1.8870 12.4357 -16.2665 14.7335age_centered2 9285 0 63.1927 108.7 0.0710 264.6chronicdiseaseHH 9285 0 0.4929 0.8051 0 1.0000activityhamperedHH 9285 0 0.3678 0.7766 0 1.0000uppereducHH 9285 0 0.5429 0.8022 0 1.0000nbchildren 9285 0 0.9117 1.6225 0 6.0000nbadultsHH 9285 0 2.2395 1.4165 1.0000 6.0000nbemployedHH 9285 0 1.6796 1.2261 0 5.0000GDPhabnuts 9285 0 27.4000 0 27.4000 27.4000unempratenuts 9285 0 10.2000 0 10.2000 10.2000

2005 EE 2648 pov_indicator 2648 0 0.6987 0.1792 0 1.0000woman 2648 0 0.5190 0.1952 0 1.0000age_centered 2648 0 -0.8065 3.5205 -16.2665 14.7335age_centered2 2648 0 81.8660 28.6931 0.0710 264.6chronicdiseaseHH 2648 0 0.4745 0.1951 0 1.0000activityhamperedHH 2648 0 0.4575 0.1946 0 1.0000uppereducHH 2648 0 0.4219 0.1929 0 1.0000nbchildren 2648 0 0.8302 0.3793 0 8.0000nbadultsHH 2648 0 2.4462 0.4073 1.0000 8.0000nbemployedHH 2648 0 1.6159 0.3584 0 6.0000GDPhabnuts 2648 0 13.7000 0 13.7000 13.7000unempratenuts 2648 0 7.9000 0 7.9000 7.9000

2006 EE 4257 pov_indicator 4257 0 0.6417 0.1471 0 1.0000woman 4257 0 0.5192 0.1533 0 1.0000age_centered 4257 0 -0.8308 2.7528 -16.2665 14.7335age_centered2 4257 0 81.1909 22.5456 0.0710 264.6chronicdiseaseHH 4257 0 0.4938 0.1534 0 1.0000activityhamperedHH 4257 0 0.4280 0.1518 0 1.0000uppereducHH 4227 30 0.4447 0.1525 0 1.0000nbchildren 4257 0 0.7761 0.2906 0 9.0000nbadultsHH 4257 0 2.4614 0.3270 1.0000 8.0000nbemployedHH 4257 0 1.6337 0.2754 0 6.0000GDPhabnuts 4257 0 15.4000 0 15.4000 15.4000unempratenuts 4257 0 5.9000 0 5.9000 5.9000

2005 IE 2872 pov_indicator 2872 0 0.0824 0.1751 0 1.0000woman 2872 0 0.5097 0.3183 0 1.0000age_centered 2872 0 -0.2720 5.5261 -16.2665 13.7335age_centered2 2872 0 75.3925 44.0678 0.0710 264.6chronicdiseaseHH 2872 0 0.3453 0.3027 0 1.0000activityhamperedHH 2872 0 0.2946 0.2903 0 1.0000uppereducHH 2777 95 0.4414 0.3168 0 1.0000nbchildren 2872 0 1.1631 0.7583 0 9.0000nbadultsHH 2872 0 2.5971 0.7238 1.0000 9.0000nbemployedHH 2872 0 1.6651 0.6233 0 6.0000GDPhabnuts 2872 0 32.4000 0 32.4000 32.4000unempratenuts 2872 0 4.3000 0 4.3000 4.3000

2006 IE 1992 pov_indicator 1992 0 0.0811 0.2023 0 1.0000woman 1992 0 0.5266 0.3700 0 1.0000age_centered 1992 0 0.5228 6.2803 -16.2665 13.7335age_centered2 1992 0 72.0837 50.2126 0.0710 264.6

chronicdiseaseHH 1992 0 0.3667 0.3571 0 1.0000activityhamperedHH 1992 0 0.2901 0.3363 0 1.0000uppereducHH 1929 63 0.4180 0.3660 0 1.0000nbchildren 1992 0 1.2204 0.9358 0 9.0000nbadultsHH 1992 0 2.5523 0.7821 1.0000 7.0000nbemployedHH 1992 0 1.5983 0.6942 0 5.0000GDPhabnuts 1992 0 34.8000 0 34.8000 34.8000unempratenuts 1992 0 4.4000 0 4.4000 4.4000

2005 EL 4642 pov_indicator 4642 0 0.2368 0.3385 0 1.0000woman 4642 0 0.5011 0.3981 0 1.0000age_centered 4642 0 -1.1845 6.9457 -16.2665 14.7335age_centered2 4642 0 77.5079 57.6718 0.0710 264.6chronicdiseaseHH 4642 0 0.2578 0.3483 0 1.0000activityhamperedHH 4642 0 0.2381 0.3391 0 1.0000uppereducHH 4446 196 0.3357 0.3777 0 1.0000nbchildren 4642 0 0.7773 0.7372 0 8.0000nbadultsHH 4642 0 2.6662 0.7512 1.0000 10.0000nbemployedHH 4642 0 1.5917 0.6483 0 7.0000

GDPhabnuts 4642 0 20.8883 4.4502 16.2000 28.3000unempratenuts 4642 0 9.9011 0.9685 8.2000 11.4000


26/34

24

2006 EL 4211 pov_indicator 4211 0 0.2574 0.3462 0 1.0000woman 4211 0 0.5024 0.3960 0 1.0000age_centered 4211 0 -0.8390 6.8965 -16.2665 14.7335age_centered2 4211 0 76.5434 57.1113 0.0710 264.6chronicdiseaseHH 4211 0 0.2626 0.3485 0 1.0000activityhamperedHH 4211 0 0.2301 0.3333 0 1.0000uppereducHH 4027 184 0.3405 0.3772 0 1.0000nbchildren 4211 0 0.7931 0.7434 0 8.0000nbadultsHH 4211 0 2.6953 0.7519 1.0000 10.0000

nbemployedHH 4211 0 1.6258 0.6779 0 7.0000GDPhabnuts 4211 0 21.8582 4.8696 16.9000 30.5000unempratenuts 4211 0 8.9479 0.5389 7.9000 9.7000

2005 ES 8429 pov_indicator 8429 0 0.2004 0.5233 0 1.0000woman 8429 0 0.4927 0.6537 0 1.0000age_centered 8429 0 -1.7718 11.3422 -16.2665 14.7335age_centered2 8429 0 78.3910 95.7980 0.0710 264.6chronicdiseaseHH 8429 0 0.3623 0.6285 0 1.0000activityhamperedHH 8429 0 0.3263 0.6131 0 1.0000uppereducHH 7794 635 0.4688 0.6525 0 1.0000nbchildren 8429 0 0.7216 1.2211 0 7.0000nbadultsHH 8429 0 2.6879 1.4681 1.0000 8.0000nbemployedHH 8429 0 1.6190 1.1830 0 6.0000GDPhabnuts 8429 0 23.1034 6.0902 15.6000 29.9000unempratenuts 8429 0 9.2296 3.6424 5.6000 19.7000

2006 ES 11438 pov_indicator 11438 0 0.1913 0.4309 0 1.0000woman 11438 0 0.4098 0.5387 0 1.0000age_centered 11438 0 -1.8626 9.4802 -16.2665 14.7335age_centered2 11438 0 78.3685 80.1413 0.0710 264.6chronicdiseaseHH 11438 0 0.3500 0.5225 0 1.0000activityhamperedHH 11438 0 0.3440 0.5204 0 1.0000uppereducHH 10605 833 0.4680 0.5469 0 1.0000nbchildren 11438 0 0.7259 0.9986 0 7.0000nbadultsHH 11438 0 2.6490 1.1568 1.0000 11.0000nbemployedHH 11438 0 1.6282 0.9546 0 7.0000GDPhabnuts 11438 0 24.6383 5.4181 16.7000 32.1000unempratenuts 11438 0 8.6267 2.7513 5.3000 21.0000

2005 FR 5876 pov_indicator 5876 0 0.0798 0.4672 0 1.0000woman 5876 0 0.5252 0.8610 0 1.0000age_centered 5876 0 1.3631 14.0051 -16.2665 14.7335age_centered2 5876 0 67.8346 116.5 0.0710 264.6chronicdiseaseHH 5876 0 0.4439 0.8567 0 1.0000activityhamperedHH 5876 0 0.2686 0.7642 0 1.0000uppereducHH 5699 177 0.3917 0.8420 0 1.0000nbchildren 5876 0 1.0526 1.9390 0 7.0000nbadultsHH 5876 0 2.2903 1.5338 1.0000 7.0000nbemployedHH 5874 2 1.5306 1.2917 0 4.0000GDPhabnuts 5876 0 25.2810 11.6435 19.6000 38.7000unempratenuts 5876 0 8.8630 3.0632 6.4000 13.2000

2006 FR 7324 pov_indicator 7324 0 0.0917 0.4396 0 1.0000woman 7324 0 0.5262 0.7608 0 1.0000age_centered 7324 0 1.2819 12.4687 -16.2665 14.7335age_centered2 7324 0 68.6154 104.2 0.0710 264.6chronicdiseaseHH 7324 0 0.4242 0.7530 0 1.0000activityhamperedHH 7324 0 0.2657 0.6730 0 1.0000uppereducHH 7119 205 0.3899 0.7433 0 1.0000nbchildren 7324 0 1.0192 1.6985 0 7.0000nbadultsHH 7324 0 2.2675 1.3719 1.0000 7.0000nbemployedHH 7324 0 1.5391 1.1152 0 4.0000GDPhabnuts 7324 0 26.1066 10.5634 20.3000 40.1000unempratenuts 7324 0 8.8601 2.5799 6.1000 12.4000

2005 IT 12955 pov_indicator 12955 0 0.1430 0.4209 0 1.0000woman 12955 0 0.4996 0.6012 0 1.0000age_centered 12955 0 -0.6673 10.2847 -16.2665 14.7335age_centered2 12955 0 73.6183 83.8161 0.0710 264.6chronicdiseaseHH 12955 0 0.2886 0.5448 0 1.0000activityhamperedHH 12955 0 0.2261 0.5030 0 1.0000uppereducHH 12797 158 0.2134 0.4925 0 1.0000nbchildren 12955 0 0.7360 1.0747 0 5.0000nbadultsHH 12955 0 2.5511 1.2129 1.0000 7.0000nbemployedHH 12955 0 1.4667 0.9832 0 5.0000GDPhabnuts 12955 0 23.6233 7.0590 15.6000 28.8000unempratenuts 12955 0 8.1620 5.5057 4.0000 15.3000

2006 IT 18048 pov_indicator 18048 0 0.1718 0.3850 0 1.0000woman 18048 0 0.4995 0.5103 0 1.0000age_centered 18048 0 -0.4297 8.6658 -16.2665 14.7335age_centered2 18048 0 72.2801 70.6271 0.0710 264.6chronicdiseaseHH 18048 0 0.2796 0.4581 0 1.0000activityhamperedHH 18048 0 0.2694 0.4528 0 1.0000

uppereducHH 17906 142 0.2191 0.4222 0 1.0000nbchildren 18048 0 0.7350 0.9174 0 6.0000nbadultsHH 18048 0 2.5369 1.0417 1.0000 7.0000nbemployedHH 18048 0 1.4865 0.8410 0 6.0000


27/34

25

GDPhabnuts 18048 0 24.5302 6.1430 16.3000 29.8000unempratenuts 18048 0 7.1662 3.8868 3.6000 12.7000

2005 CY 2357 pov_indicator 2357 0 0.0719 0.0821 0 1.0000woman 2357 0 0.5104 0.1588 0 1.0000age_centered 2357 0 -1.1817 2.8284 -16.2665 14.7335age_centered2 2357 0 80.6322 23.5761 0.0710 264.6chronicdiseaseHH 2357 0 0.3778 0.1541 0 1.0000activityhamperedHH 2357 0 0.3540 0.1520 0 1.0000

uppereducHH 2315 42 0.4840 0.1589 0 1.0000nbchildren 2357 0 0.9912 0.3387 0 8.0000nbadultsHH 2357 0 2.7608 0.3348 1.0000 7.0000nbemployedHH 2357 0 1.8203 0.2727 0 6.0000GDPhabnuts 2357 0 20.4000 0 20.4000 20.4000unempratenuts 2357 0 5.3000 0 5.3000 5.3000

2006 CY 3373 pov_indicator 3373 0 0.0552 0.0601 0 1.0000woman 3373 0 0.5142 0.1316 0 1.0000age_centered 3373 0 -1.0617 2.3468 -16.2665 14.7335age_centered2 3373 0 80.5458 19.5748 0.0710 264.6chronicdiseaseHH 3373 0 0.4170 0.1298 0 1.0000activityhamperedHH 3373 0 0.2706 0.1170 0 1.0000uppereducHH 3319 54 0.4900 0.1317 0 1.0000nbchildren 3373 0 1.0275 0.2843 0 8.0000nbadultsHH 3373 0 2.8076 0.2898 1.0000 7.0000nbemployedHH 3373 0 1.8817 0.2335 0 6.0000GDPhabnuts 3373 0 21.3000 0 21.3000 21.3000unempratenuts 3373 0 4.5000 0 4.5000 4.5000

2005 LV 2704 pov_indicator 2704 0 0.8089 0.2019 0 1.0000woman 2704 0 0.5195 0.2566 0 1.0000age_centered 2704 0 -0.6736 4.5409 -16.2665 14.7335age_centered2 2704 0 78.6490 37.5947 0.0710 264.6chronicdiseaseHH 2704 0 0.5020 0.2568 0 1.0000activityhamperedHH 2704 0 0.4826 0.2566 0 1.0000uppereducHH 2686 18 0.3015 0.2357 0 1.0000nbchildren 2704 0 0.7461 0.4546 0 7.0000nbadultsHH 2704 0 2.5481 0.5490 1.0000 6.0000nbemployedHH 2704 0 1.6262 0.4794 0 5.0000GDPhabnuts 2704 0 10.9000 0 10.9000 10.9000unempratenuts 2704 0 8.9000 0 8.9000 8.9000

2006 LV 3438 pov_indicator 3438 0 0.7396 0.1993 0 1.0000woman 3438 0 0.5180 0.2270 0 1.0000age_centered 3438 0 -0.7024 4.0261 -16.2665 14.7335age_centered2 3438 0 79.0465 32.8692 0.0710 264.6chronicdiseaseHH 3438 0 0.5185 0.2270 0 1.0000activityhamperedHH 3438 0 0.4850 0.2270 0 1.0000uppereducHH 3417 21 0.3494 0.2165 0 1.0000nbchildren 3438 0 0.7660 0.4175 0 11.0000nbadultsHH 3438 0 2.6926 0.4952 1.0000 6.0000nbemployedHH 3438 0 1.7491 0.4339 0 5.0000GDPhabnuts 3438 0 12.4000 0 12.4000 12.4000unempratenuts 3438 0 6.8000 0 6.8000 6.8000

2005 LT 2527 pov_indicator 2527 0 0.7906 0.2636 0 1.0000woman 2527 0 0.5165 0.3237 0 1.0000age_centered 2527 0 -0.8742 5.5828 -16.2665 14.7335age_centered2 2527 0 75.0490 46.8519 0.0710 264.6chronicdiseaseHH 2527 0 0.3968 0.3169 0 1.0000activityhamperedHH 2527 0 0.3962 0.3168 0 1.0000uppereducHH 2455 72 0.3853 0.3162 0 1.0000nbchildren 2527 0 0.9105 0.6330 0 6.0000nbadultsHH 2527 0 2.5087 0.6445 1.0000 9.0000nbemployedHH 2527 0 1.6172 0.5482 0 5.0000

GDPhabnuts 2527 0 11.9000 0 11.9000 11.9000unempratenuts 2527 0 8.3000 0 8.3000 8.3000

2006 LT 3713 pov_indicator 3713 0 0.7549 0.2280 0 1.0000woman 3713 0 0.5173 0.2649 0 1.0000age_centered 3713 0 -0.6671 4.5500 -16.2665 14.7335age_centered2 3713 0 74.1260 37.1506 0.0710 264.6chronicdiseaseHH 3713 0 0.4288 0.2623 0 1.0000activityhamperedHH 3713 0 0.3475 0.2524 0 1.0000uppereducHH 3613 100 0.4118 0.2618 0 1.0000nbchildren 3713 0 0.9010 0.5201 0 7.0000nbadultsHH 3713 0 2.5249 0.5195 1.0000 9.0000nbemployedHH 3713 0 1.6651 0.4300 0 5.0000GDPhabnuts 3713 0 13.1000 0 13.1000 13.1000unempratenuts 3713 0 5.6000 0 5.6000 5.6000

2005 LU 4322 pov_indicator 4322 0 0.0106 0.0191 0 1.0000woman 4322 0 0.4941 0.0930 0 1.0000age_centered 4322 0 -0.5958 1.5620 -16.2665 14.7335

age_centered2 4322 0 70.9201 12.6713 0.0710 264.6chronicdiseaseHH 4322 0 0.3231 0.0870 0 1.0000activityhamperedHH 4322 0 0.3213 0.0868 0 1.0000uppereducHH 4210 112 0.3544 0.0895 0 1.0000


28/34

26

nbchildren 4322 0 0.9074 0.2003 0 7.0000nbadultsHH 4322 0 2.3891 0.1848 1.0000 7.0000nbemployedHH 4322 0 1.5991 0.1397 0 5.0000GDPhabnuts 4322 0 57.1000 0 57.1000 57.1000unempratenuts 4322 0 4.5000 0 4.5000 4.5000

2006 LU 4601 pov_indicator 4601 0 0.0192 0.0251 0 1.0000woman 4601 0 0.4979 0.0917 0 1.0000age_centered 4601 0 -0.8068 1.5614 -16.2665 14.7335

age_centered2 4601 0 73.1864 12.9800 0.0710 264.6chronicdiseaseHH 4601 0 0.3346 0.0865 0 1.0000activityhamperedHH 4601 0 0.3191 0.0855 0 1.0000uppereducHH 4494 107 0.3524 0.0882 0 1.0000nbchildren 4601 0 0.9019 0.1980 0 6.0000nbadultsHH 4601 0 2.3624 0.1804 1.0000 6.0000nbemployedHH 4601 0 1.5974 0.1353 0 5.0000GDPhabnuts 4601 0 63.1000 0 63.1000 63.1000unempratenuts 4601 0 4.7000 0 4.7000 4.7000

2005 HU 4353 pov_indicator 4353 0 0.7286 0.3793 0 1.0000woman 4353 0 0.5111 0.4264 0 1.0000age_centered 4353 0 -0.7273 7.8870 -16.2665 14.7335age_centered2 4353 0 86.0190 62.2732 0.0710 264.6chronicdiseaseHH 4353 0 0.5287 0.4258 0 1.0000activityhamperedHH 4353 0 0.4448 0.4239 0 1.0000uppereducHH 4350 3 0.2556 0.3721 0 1.0000nbchildren 4353 0 0.8801 0.9083 0 7.0000nbadultsHH 4353 0 2.6492 0.8968 1.0000 9.0000nbemployedHH 4353 0 1.5751 0.7810 0 5.0000GDPhabnuts 4353 0 14.2595 4.9600 9.3000 23.2000unempratenuts 4353 0 7.3201 1.4527 5.1000 9.2000

2006 HU 6609 pov_indicator 6609 0 0.6904 0.3197 0 1.0000woman 6609 0 0.5155 0.3456 0 1.0000age_centered 6609 0 -0.7190 6.3795 -16.2665 14.7335age_centered2 6609 0 85.6337 50.2613 0.0710 264.6chronicdiseaseHH 6609 0 0.4654 0.3449 0 1.0000activityhamperedHH 6609 0 0.3874 0.3369 0 1.0000uppereducHH 6609 0 0.2955 0.3155 0 1.0000nbchildren 6609 0 0.8703 0.7353 0 8.0000nbadultsHH 6609 0 2.6738 0.7254 1.0000 7.0000nbemployedHH 6609 0 1.5826 0.6395 0 5.0000GDPhabnuts 6609 0 15.0909 4.4141 9.7000 24.9000unempratenuts 6609 0 7.5619 1.4049 5.1000 9.9000

2005 NL 8574 pov_indicator 8574 0 0.0581 0.1868 0 1.0000woman 8574 0 0.4921 0.3992 0 1.0000age_centered 8574 0 -0.4148 6.8706 -16.2665 14.7335age_centered2 8574 0 74.2244 56.6392 0.0710 264.6chronicdiseaseHH 8574 0 0.2497 0.3456 0 1.0000activityhamperedHH 8574 0 0.1563 0.2899 0 1.0000uppereducHH 8206 368 0.4600 0.3967 0 1.0000nbchildren 8574 0 0.9032 0.8799 0 6.0000nbadultsHH 8574 0 2.1546 0.6870 1.0000 7.0000nbemployedHH 8407 167 1.4760 0.6251 0 5.0000GDPhabnuts 8574 0 29.4000 0 29.4000 29.4000unempratenuts 8574 0 4.7000 0 4.7000 4.7000

2006 NL 9287 pov_indicator 9287 0 0.0318 0.1310 0 1.0000woman 9287 0 0.4975 0.3732 0 1.0000age_centered 9287 0 -0.2668 6.3946 -16.2665 14.7335age_centered2 9287 0 73.4840 53.0152 0.0710 264.6chronicdiseaseHH 9287 0 0.2542 0.3249 0 1.0000activityhamperedHH 9287 0 0.1644 0.2766 0 1.0000uppereducHH 8949 338 0.4702 0.3716 0 1.0000

nbchildren 9287 0 0.9265 0.8325 0 6.0000nbadultsHH 9287 0 2.1678 0.6369 1.0000 7.0000nbemployedHH 9279 8 1.6542 0.5648 0 5.0000GDPhabnuts 9287 0 30.9000 0 30.9000 30.9000unempratenuts 9287 0 3.9000 0 3.9000 3.9000

2005 AT 3882 pov_indicator 3882 0 0.0522 0.1828 0 1.0000woman 3882 0 0.5056 0.4107 0 1.0000age_centered 3882 0 -0.1534 6.9123 -16.2665 14.7335age_centered2 3882 0 70.8171 57.3363 0.0710 264.6chronicdiseaseHH 3882 0 0.2931 0.3739 0 1.0000activityhamperedHH 3882 0 0.3432 0.3900 0 1.0000uppereducHH 3882 0 0.3181 0.3826 0 1.0000nbchildren 3882 0 0.8440 0.8552 0 6.0000nbadultsHH 3882 0 2.3847 0.8766 1.0000 8.0000nbemployedHH 3882 0 1.6527 0.7373 0 7.0000GDPhabnuts 3882 0 28.0319 1.8223 23.9000 29.6000unempratenuts 3882 0 5.1660 1.0814 3.9000 6.7000

2006 AT 5469 pov_indicator 5469 0 0.0555 0.1577 0 1.0000woman 5469 0 0.5080 0.3443 0 1.0000age_centered 5469 0 -0.3177 5.7667 -16.2665 14.7335age_centered2 5469 0 70.2327 48.1919 0.0710 264.6


29/34

27

chronicdiseaseHH 5469 0 0.2754 0.3076 0 1.0000activityhamperedHH 5469 0 0.3202 0.3213 0 1.0000uppereducHH 5469 0 0.3149 0.3198 0 1.0000nbchildren 5469 0 0.8224 0.7036 0 6.0000nbadultsHH 5469 0 2.3748 0.7198 1.0000 8.0000nbemployedHH 5469 0 1.6395 0.6173 0 6.0000GDPhabnuts 5469 0 29.4096 1.4577 25.3000 30.8000unempratenuts 5469 0 4.7459 0.9453 3.3000 6.3000

2005 PL 10627 pov_indicator 10627 0 0.7381 0.4729 0 1.0000woman 10627 0 0.5039 0.5378 0 1.0000age_centered 10627 0 -0.6632 9.9424 -16.2665 14.7335age_centered2 10627 0 85.8863 79.9035 0.0710 264.6chronicdiseaseHH 10627 0 0.4761 0.5372 0 1.0000activityhamperedHH 10627 0 0.2527 0.4674 0 1.0000uppereducHH 10561 66 0.2646 0.4748 0 1.0000nbchildren 10627 0 0.8989 1.1369 0 11.0000nbadultsHH 10627 0 2.7400 1.3407 1.0000 9.0000nbemployedHH 10564 63 1.4003 1.0024 0 5.0000GDPhabnuts 10627 0 11.5603 2.5551 8.2000 15.7000unempratenuts 10627 0 17.7915 2.0488 15.7000 21.4000

2006 PL 14644 pov_indicator 14644 0 0.7258 0.4102 0 1.0000woman 14644 0 0.5037 0.4598 0 1.0000age_centered 14644 0 -0.6473 8.4644 -16.2665 14.7335age_centered2 14644 0 85.1495 67.2797 0.0710 264.6chronicdiseaseHH 14644 0 0.4808 0.4594 0 1.0000activityhamperedHH 14644 0 0.3260 0.4310 0 1.0000uppereducHH 12928 1716 0.2834 0.4143 0 1.0000nbchildren 14644 0 0.9003 0.9688 0 11.0000nbadultsHH 14644 0 2.9673 1.1522 1.0000 9.0000nbemployedHH 13001 1643 1.5990 0.8760 0 6.0000GDPhabnuts 14644 0 12.3506 2.4094 8.7000 17.0000unempratenuts 14644 0 13.9915 1.0485 12.7000 16.4000

2005 PT 2671 pov_indicator 2671 0 0.4168 0.5441 0 1.0000woman 2671 0 0.5072 0.5517 0 1.0000age_centered 2671 0 -0.6899 9.8001 -16.2665 14.7335age_centered2 2671 0 79.3436 79.7975 0.0710 264.6chronicdiseaseHH 2671 0 0.4695 0.5507 0 1.0000activityhamperedHH 2671 0 0.4113 0.5430 0 1.0000uppereducHH 2198 473 0.2303 0.4696 0 1.0000nbchildren 2671 0 0.8315 0.9949 0 9.0000nbadultsHH 2671 0 2.8063 1.1722 1.0000 8.0000nbemployedHH 2671 0 1.8021 1.0142 0 7.0000GDPhabnuts 2671 0 17.3000 0 17.3000 17.3000unempratenuts 2671 0 7.6000 0 7.6000 7.6000

2006 PT 3662 pov_indicator 3662 0 0.4019 0.4575 0 1.0000woman 3662 0 0.5112 0.4665 0 1.0000age_centered 3662 0 -0.3565 8.3039 -16.2665 14.7335age_centered2 3662 0 79.3109 66.5722 0.0710 264.6chronicdiseaseHH 3662 0 0.4380 0.4630 0 1.0000activityhamperedHH 3662 0 0.3936 0.4559 0 1.0000uppereducHH 3049 613 0.2419 0.4037 0 1.0000nbchildren 3662 0 0.7832 0.8223 0 9.0000nbadultsHH 3662 0 2.8086 1.0182 1.0000 9.0000nbemployedHH 3662 0 1.8287 0.8564 0 7.0000GDPhabnuts 3662 0 18.0000 0 18.0000 18.0000unempratenuts 3662 0 7.7000 0 7.7000 7.7000

2005 SI 8671 pov_indicator 8671 0 0.1266 0.0915 0 1.0000woman 8667 4 0.4899 0.1376 0 1.0000age_centered 8671 0 -0.6833 2.4495 -16.2665 14.7335

age_centered2 8671 0 79.6898 19.9780 0.0710 264.6chronicdiseaseHH 8671 0 0.2501 0.1192 0 1.0000activityhamperedHH 8671 0 0.2177 0.1136 0 1.0000uppereducHH 8471 200 0.2149 0.1128 0 1.0000nbchildren 8671 0 0.7566 0.2591 0 9.0000nbadultsHH 8671 0 2.9380 0.3047 1.0000 9.0000nbemployedHH 8658 13 1.7535 0.2551 0 6.0000GDPhabnuts 8671 0 19.6000 0 19.6000 19.6000unempratenuts 8671 0 6.5000 0 6.5000 6.5000

2006 SI 11271 pov_indicator 11271 0 0.1306 0.0808 0 1.0000woman 11263 8 0.4929 0.1200 0 1.0000age_centered 11271 0 -0.7373 2.1167 -16.2665 14.7335age_centered2 11271 0 78.4043 17.6044 0.0710 264.6chronicdiseaseHH 11271 0 0.2915 0.1090 0 1.0000activityhamperedHH 11271 0 0.2126 0.0981 0 1.0000uppereducHH 11189 82 0.2549 0.1046 0 1.0000nbchildren 11271 0 0.7634 0.2254 0 9.0000nbadultsHH 11271 0 2.9358 0.2660 1.0000 9.0000

nbemployedHH 11262 9 1.7851 0.2174 0 6.0000GDPhabnuts 11271 0 20.7000 0 20.7000 20.7000unempratenuts 11271 0 6.0000 0 6.0000 6.0000


30/34

28

2005 SK 3447 pov_indicator 3447 0 0.7661 0.3005 0 1.0000woman 3447 0 0.5198 0.3547 0 1.0000age_centered 3447 0 0.1174 6.4023 -16.2665 14.7335age_centered2 3447 0 81.3319 53.0850 0.0710 264.6chronicdiseaseHH 3447 0 0.4441 0.3528 0 1.0000activityhamperedHH 3447 0 0.3993 0.3477 0 1.0000uppereducHH 3432 15 0.3085 0.3279 0 1.0000nbchildren 3447 0 0.8904 0.7283 0 6.0000nbadultsHH 3447 0 3.0942 0.8632 1.0000 8.0000

nbemployedHH 3447 0 1.9700 0.7508 0 8.0000GDPhabnuts 3447 0 13.5000 0 13.5000 13.5000unempratenuts 3447 0 16.3000 0 16.3000 16.3000

2006 SK 4898 pov_indicator 4898 0 0.7157 0.2680 0 1.0000woman 4898 0 0.5217 0.2968 0 1.0000age_centered 4898 0 0.0502 5.3919 -16.2665 14.7335age_centered2 4898 0 82.3628 44.6004 0.0710 264.6chronicdiseaseHH 4898 0 0.4327 0.2944 0 1.0000activityhamperedHH 4898 0 0.4442 0.2952 0 1.0000uppereducHH 4870 28 0.3248 0.2783 0 1.0000nbchildren 4898 0 0.8442 0.5925 0 8.0000nbadultsHH 4898 0 3.1409 0.7288 1.0000 8.0000nbemployedHH 4898 0 2.0315 0.6170 0 8.0000GDPhabnuts 4898 0 15.0000 0 15.0000 15.0000unempratenuts 4898 0 13.4000 0 13.4000 13.4000

2005 FI 4021 pov_indicator 4021 0 0.0529 0.1404 0 1.0000woman 4021 0 0.4953 0.3137 0 1.0000age_centered 4021 0 -0.3310 5.6323 -16.2665 14.7335age_centered2 4021 0 80.6732 46.9628 0.0710 264.6chronicdiseaseHH 4021 0 0.3076 0.2896 0 1.0000activityhamperedHH 4021 0 0.2930 0.2856 0 1.0000uppereducHH 3927 94 0.5226 0.3133 0 1.0000nbchildren 4021 0 0.8869 0.7442 0 10.0000nbadultsHH 4021 0 2.1009 0.5226 1.0000 7.0000nbemployedHH 4018 3 1.4707 0.4894 0 5.0000GDPhabnuts 4021 0 25.6182 2.5938 19.1000 29.6000unempratenuts 4021 0 8.5095 1.1694 6.9000 11.7000

2006 FI 5528 pov_indicator 5528 0 0.0622 0.1290 0 1.0000woman 5528 0 0.4953 0.2669 0 1.0000age_centered 5528 0 -0.3237 4.7976 -16.2665 14.7335age_centered2 5528 0 80.8606 39.6984 0.0710 264.6chronicdiseaseHH 5528 0 0.2901 0.2423 0 1.0000activityhamperedHH 5528 0 0.2894 0.2421 0 1.0000uppereducHH 5404 124 0.5179 0.2668 0 1.0000nbchildren 5528 0 0.8898 0.6325 0 12.0000nbadultsHH 5528 0 2.0887 0.4440 1.0000 8.0000nbemployedHH 5528 0 1.4683 0.3940 0 4.0000GDPhabnuts 5528 0 27.1434 2.3172 20.2000 31.3000unempratenuts 5528 0 7.7955 0.9975 6.3000 11.3000

2005 SE 3541 pov_indicator 3541 0 0.0452 0.1750 0 1.0000woman 3541 0 0.5016 0.4211 0 1.0000age_centered 3541 0 -0.3763 7.4383 -16.2665 14.7335age_centered2 3541 0 78.1345 60.4577 0.0710 264.6chronicdiseaseHH 3541 0 0.2827 0.3793 0 1.0000activityhamperedHH 3541 0 0.1482 0.2992 0 1.0000uppereducHH 3414 127 0.4394 0.4189 0 1.0000nbchildren 3541 0 0.9789 0.9351 0 8.0000nbadultsHH 3541 0 2.1022 0.6935 1.0000 7.0000nbemployedHH 3155 386 1.5457 0.5973 0 5.0000GDPhabnuts 3541 0 27.1000 0 27.1000 27.1000unempratenuts 3541 0 7.5000 0 7.5000 7.5000

2006 SE 4751 pov_indicator 4751 0 0.0527 0.1577 0 1.0000woman 4751 0 0.5011 0.3531 0 1.0000age_centered 4751 0 -0.4989 6.1470 -16.2665 14.7335age_centered2 4751 0 76.0001 49.7601 0.0710 264.6chronicdiseaseHH 4751 0 0.2920 0.3211 0 1.0000activityhamperedHH 4751 0 0.1593 0.2585 0 1.0000uppereducHH 4534 217 0.4728 0.3530 0 1.0000nbchildren 4751 0 1.0203 0.7818 0 6.0000nbadultsHH 4751 0 2.0672 0.5636 1.0000 7.0000nbemployedHH 4210 541 1.5782 0.4843 0 5.0000GDPhabnuts 4751 0 28.7000 0 28.7000 28.7000unempratenuts 4751 0 7.1000 0 7.1000 7.1000

2005 UK 6224 pov_indicator 6224 0 0.0739 0.5272 0 1.0000woman 6224 0 0.5260 1.0064 0 1.0000age_centered 6224 0 -1.9273 16.1286 -16.2665 13.7335age_centered2 6224 0 67.7453 142.1 0.0710 264.6chronicdiseaseHH 6224 0 0.4031 0.9887 0 1.0000activityhamperedHH 6224 0 0.2293 0.8473 0 1.0000

uppereducHH 4949 1275 0.5177 1.0064 0 1.0000nbchildren 6224 0 1.2542 2.3990 0 6.0000nbadultsHH 6224 0 2.1928 1.7624 1.0000 7.0000nbemployedHH 4766 1458 1.5478 1.7058 0 6.0000


31/34

29

GDPhabnuts 6224 0 27.4000 0 27.4000 27.4000unempratenuts 6224 0 4.8000 0 4.8000 4.8000

2006 UK 7594 pov_indicator 7594 0 0.0837 0.4988 0 1.0000woman 7594 0 0.5363 0.8983 0 1.0000age_centered 7594 0 -1.5980 14.5325 -16.2665 13.7335age_centered2 7594 0 67.6428 128.3 0.0710 264.6chronicdiseaseHH 7594 0 0.4049 0.8842 0 1.0000activityhamperedHH 7594 0 0.2199 0.7461 0 1.0000

uppereducHH 5998 1596 0.5080 0.8985 0 1.0000nbchildren 7594 0 1.2305 2.1147 0 6.0000nbadultsHH 7594 0 2.1941 1.5651 1.0000 8.0000nbemployedHH 6118 1476 1.5570 1.4788 0 5.0000GDPhabnuts 7594 0 28.4000 0 28.4000 28.4000unempratenuts 7594 0 5.4000 0 5.4000 5.4000

2005 IS 1583 pov_indicator 1583 0 0.0333 0.0418 0 1.0000woman 1583 0 0.5066 0.1164 0 1.0000age_centered 1583 0 -1.0735 2.0227 -16.2665 14.7335age_centered2 1583 0 76.5700 16.9520 0.0710 264.6chronicdiseaseHH 1583 0 0.2143 0.0956 0 1.0000activityhamperedHH 1583 0 0.1521 0.0836 0 1.0000uppereducHH 1359 224 0.3835 0.1150 0 1.0000nbchildren 1583 0 1.1989 0.2545 0 5.0000nbadultsHH 1583 0 2.4249 0.2303 1.0000 7.0000nbemployedHH 1318 265 1.8715 0.1947 0 6.0000GDPhabnuts 1583 0 29.3000 0 29.3000 29.3000unempratenuts 1583 0 2.5000 0 2.5000 2.5000

2006 IS 2240 pov_indicator 2240 0 0.0322 0.0348 0 1.0000woman 2240 0 0.4993 0.0986 0 1.0000age_centered 2240 0 -0.9177 1.7342 -16.2665 14.7335age_centered2 2240 0 78.2001 14.4429 0.0710 264.6chronicdiseaseHH 2240 0 0.2180 0.0814 0 1.0000activityhamperedHH 2240 0 0.1579 0.0719 0 1.0000uppereducHH 2021 219 0.4147 0.0983 0 1.0000nbchildren 2240 0 1.1305 0.2161 0 5.0000nbadultsHH 2240 0 2.3563 0.1875 1.0000 7.0000nbemployedHH 2054 186 1.7410 0.1499 0 5.0000GDPhabnuts 2240 0 29.3000 0 29.3000 29.3000unempratenuts 2240 0 2.8000 0 2.8000 2.8000

2005 NO 4088 pov_indicator 4088 0 0.0326 0.1001 0 1.0000woman 4088 0 0.4919 0.2816 0 1.0000age_centered 4088 0 -0.6817 4.8555 -16.2665 14.7335age_centered2 4088 0 74.7612 40.0863 0.0710 264.6chronicdiseaseHH 4088 0 0.2545 0.2454 0 1.0000activityhamperedHH 4088 0 0.1610 0.2070 0 1.0000uppereducHH 3887 201 0.3970 0.2753 0 1.0000nbchildren 4088 0 1.0059 0.6332 0 8.0000nbadultsHH 4088 0 2.0123 0.4476 1.0000 8.0000nbemployedHH 3996 92 1.4737 0.4073 0 4.0000GDPhabnuts 4088 0 39.6000 0 39.6000 39.6000unempratenuts 4088 0 4.4000 0 4.4000 4.4000

2006 NO 3926 pov_indicator 3926 0 0.0312 0.0978 0 1.0000woman 3926 0 0.4909 0.2814 0 1.0000age_centered 3926 0 -0.5332 4.7661 -16.2665 14.7335age_centered2 3926 0 71.9978 39.3175 0.0710 264.6chronicdiseaseHH 3926 0 0.2769 0.2518 0 1.0000activityhamperedHH 3926 0 0.1713 0.2120 0 1.0000uppereducHH 3583 343 0.4123 0.2776 0 1.0000nbchildren 3926 0 0.9872 0.6287 0 8.0000nbadultsHH 3926 0 2.0046 0.4367 1.0000 8.0000nbemployedHH 3732 194 1.5183 0.4010 0 5.0000

GDPhabnuts 3926 0 43.4000 0 43.4000 43.4000unempratenuts 3926 0 3.4000 0 3.4000 3.4000---------------------------------------------------------------------------------------------------


32/34


33/34


34/34

B.P. 48L-4501 DifferdangeTl.: +352 58.58.55-801

www.ceps.lu

Documents

CEPS Macro Determinants of Individual Income Poverty In